AuroraMiddleware/scuffed-code
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

ICU-484 reimplement LATIN_1 to work correctly with offsets and UTF-16

Browse Source 
X-SVN-Rev: 3285
This commit is contained in:
Markus Scherer 2000-12-20 01:22:02 +00:00
parent 4bd8af1a79
commit 8d9bdf7a1e
1 changed files with 250 additions and 142 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

392
icu4c/source/common/ucnvlat1.c
View File

@ -10,168 +10,276 @@
* *
* created on: 2000feb07 * created on: 2000feb07
* created by: Markus W. Scherer * created by: Markus W. Scherer
* Change history:
*
* 06/29/2000 helena Major rewrite of the callback APIs.
*/ */
#include "unicode/utypes.h" #include "unicode/utypes.h"
#include "ucmp16.h" #include "unicode/ucnv.h"
#include "ucmp8.h"
#include "unicode/ucnv_err.h" #include "unicode/ucnv_err.h"
#include "ucnv_bld.h" #include "ucnv_bld.h"
#include "unicode/ucnv.h"
#include "ucnv_cnv.h" #include "ucnv_cnv.h"
/* ISO 8859-1 --------------------------------------------------------------- */ /* ISO 8859-1 --------------------------------------------------------------- */
U_CFUNC void T_UConverter_toUnicode_LATIN_1 (UConverterToUnicodeArgs * args, /* This is a table-less and callback-less version of _MBCSSingleToBMPWithOffsets(). */
UErrorCode * err) U_CFUNC void
{ _Latin1ToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs,
unsigned char *mySource = (unsigned char *) args->source; UErrorCode *pErrorCode) {
UChar *myTarget = args->target; const uint8_t *source;
int32_t sourceLength = args->sourceLimit - (char *) mySource; UChar *target;
int32_t readLen = 0; int32_t targetCapacity, length;
int32_t i = 0; int32_t *offsets;
/*Since there is no risk of encountering illegal Chars /* set up the local pointers */
*we need to pad our latin1 chars to create Unicode codepoints source=(const uint8_t *)pArgs->source;
*we need to go as far a min(targetLen, sourceLen) target=pArgs->target;
*in case we don't have enough buffer space targetCapacity=pArgs->targetLimit-pArgs->target;
*we set the error flag accordingly
*/ /*
if ((args->targetLimit - args->target) < sourceLength) * since the conversion here is 1:1 UChar:uint8_t, we need only one counter
{ * for the minimum of the sourceLength and targetCapacity
readLen = args->targetLimit - args->target; */
*err = U_BUFFER_OVERFLOW_ERROR; length=(const uint8_t *)pArgs->sourceLimit-source;
} if(length<=targetCapacity) {
else targetCapacity=length;
{ } else {
readLen = args->sourceLimit - (char *) mySource; /* target will be full */
*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
length=targetCapacity;
} }
for (i = 0; i < readLen; i++) myTarget[i] = (UChar) mySource[i]; /* conversion loop */
while(targetCapacity>0) {
*target++=*source++;
--targetCapacity;
}
args->target += i; /* write back the updated pointers */
args->source += i; pArgs->source=(const char *)source;
return; pArgs->target=target;
/* set offsets */
offsets=pArgs->offsets;
if(offsets!=NULL) {
int32_t sourceIndex=0;
while(length>0) {
*offsets++=sourceIndex++;
--length;
}
pArgs->offsets=offsets;
}
} }
U_CFUNC void T_UConverter_fromUnicode_LATIN_1 (UConverterFromUnicodeArgs * args, /* This is a table-less and callback-less version of _MBCSSingleGetNextUChar(). */
UErrorCode * err) U_CFUNC UChar32
{ _Latin1GetNextUChar(UConverterToUnicodeArgs *pArgs,
const UChar *mySource = args->source; UErrorCode *pErrorCode) {
unsigned char *myTarget = (unsigned char *) args->target; const uint8_t *source=(const uint8_t *)pArgs->source;
int32_t mySourceIndex = 0; if(source<(const uint8_t *)pArgs->sourceLimit) {
int32_t myTargetIndex = 0; pArgs->source=(const char *)(source+1);
int32_t targetLength = args->targetLimit - (char *) myTarget; return *source;
int32_t sourceLength = args->sourceLimit - mySource; }
UConverterCallbackReason reason;
/*writing the char to the output stream */ /* no output because of empty input */
while (mySourceIndex < sourceLength) *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
{ return 0xffff;
if (myTargetIndex < targetLength) }
{
if (mySource[mySourceIndex] < 0x0100) /* This is a table-less version of _MBCSSingleFromBMPWithOffsets(). */
{ U_CFUNC void
/*writes the char to the output stream */ _Latin1FromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs,
myTarget[myTargetIndex++] = (char) mySource[mySourceIndex++]; UErrorCode *pErrorCode) {
UConverter *cnv;
const UChar *source, *sourceLimit, *lastSource;
uint8_t *target;
int32_t targetCapacity, length;
int32_t *offsets;
UChar32 c, max;
int32_t sourceIndex;
UConverterCallbackReason reason;
int32_t i;
/* set up the local pointers */
cnv=pArgs->converter;
source=pArgs->source;
sourceLimit=pArgs->sourceLimit;
target=(uint8_t *)pArgs->target;
targetCapacity=pArgs->targetLimit-pArgs->target;
offsets=pArgs->offsets;
max=0xff; /* ### 0x7f for US-ASCII */
/* get the converter state from UConverter */
c=cnv->fromUSurrogateLead;
/* sourceIndex=-1 if the current character began in the previous buffer */
sourceIndex= c==0 ? 0 : -1;
lastSource=source;
/*
* since the conversion here is 1:1 UChar:uint8_t, we need only one counter
* for the minimum of the sourceLength and targetCapacity
*/
length=sourceLimit-source;
if(length<targetCapacity) {
targetCapacity=length;
}
/* conversion loop */
if(c!=0 && targetCapacity>0) {
goto getTrail;
}
while(targetCapacity>0) {
/*
* Get a correct Unicode code point:
* a single UChar for a BMP code point or
* a matched surrogate pair for a "surrogate code point".
*/
c=*source++;
if(c<=max) {
/* convert the Unicode code point */
*target++=(uint8_t)c;
--targetCapacity;
/* normal end of conversion: prepare for a new character */
c=0;
} else {
if(!UTF_IS_SURROGATE(c)) {
/* callback(unassigned) */
reason=UCNV_UNASSIGNED;
*pErrorCode=U_INVALID_CHAR_FOUND;
} else if(UTF_IS_SURROGATE_FIRST(c)) {
getTrail:
if(source<sourceLimit) {
/* test the following code unit */
UChar trail=*source;
if(UTF_IS_SECOND_SURROGATE(trail)) {
++source;
c=UTF16_GET_PAIR_VALUE(c, trail);
/* this codepage does not map supplementary code points */
/* callback(unassigned) */
reason=UCNV_UNASSIGNED;
*pErrorCode=U_INVALID_CHAR_FOUND;
} else {
/* this is an unmatched lead code unit (1st surrogate) */
/* callback(illegal) */
reason=UCNV_ILLEGAL;
*pErrorCode=U_ILLEGAL_CHAR_FOUND;
}
} else {
/* no more input */
break;
}
} else {
/* this is an unmatched trail code unit (2nd surrogate) */
/* callback(illegal) */
reason=UCNV_ILLEGAL;
*pErrorCode=U_ILLEGAL_CHAR_FOUND;
} }
else
{
*err = U_INVALID_CHAR_FOUND;
reason = UCNV_UNASSIGNED;
args->converter->invalidUCharBuffer[0] = (UChar)mySource[mySourceIndex];
args->converter->invalidUCharLength = 1;
if (UTF_IS_LEAD(mySource[mySourceIndex++]))
{
if (mySourceIndex < sourceLength)
{
if (UTF_IS_TRAIL(mySource[mySourceIndex]))
{
args->converter->invalidUCharBuffer[1] = (UChar)mySource[mySourceIndex];
args->converter->invalidUCharLength++;
mySourceIndex++;
}
else
{
reason = UCNV_ILLEGAL;
}
}
else if (args->flush == TRUE)
{
reason = UCNV_ILLEGAL;
*err = U_TRUNCATED_CHAR_FOUND;
}
else
{
args->converter->fromUSurrogateLead = args->converter->invalidUCharBuffer[0];
/* do not call the callback */
}
}
if (args->converter->fromUSurrogateLead == 0)
{
const UChar *saveSource = args->source;
char *saveTarget = args->target;
int32_t *saveOffset = args->offsets;
/* Needed explicit cast for myTarget on MVS to make compiler happy - JJD */
args->target = (char*)myTarget + myTargetIndex;;
args->source = mySource + mySourceIndex;
FromU_CALLBACK_MACRO(args->converter->fromUContext, /* call the callback function with all the preparations and post-processing */
args, /* get the number of code units for c to correctly advance sourceIndex after the callback call */
args->converter->invalidUCharBuffer, length=UTF_CHAR_LENGTH(c);
args->converter->invalidUCharLength,
(UChar32) (args->converter->invalidUCharLength == 2 ? /* set offsets since the start or the last callback */
UTF16_GET_PAIR_VALUE(args->converter->invalidUCharBuffer[0], if(offsets!=NULL) {
args->converter->invalidUCharBuffer[1]) int32_t count=(int32_t)(source-lastSource);
: args->converter->invalidUCharBuffer[0]),
reason, /* do not set the offset for the callback-causing character */
err); count-=length;
args->source = saveSource;
args->target = saveTarget; while(count>0) {
args->offsets = saveOffset; *offsets++=sourceIndex++;
if (U_FAILURE (*err)) --count;
{ }
break; /* offset and sourceIndex are now set for the current character */
} }
args->converter->invalidUCharLength = 0;
} /* update the arguments structure */
pArgs->source=source;
pArgs->target=(char *)target;
pArgs->offsets=offsets;
/* set the converter state in UConverter to deal with the next character */
cnv->fromUSurrogateLead=0;
/* write the code point as code units */
i=0;
UTF_APPEND_CHAR_UNSAFE(cnv->invalidUCharBuffer, i, c);
cnv->invalidUCharLength=(int8_t)i;
/* i==length */
/* call the callback function */
cnv->fromUCharErrorBehaviour(cnv->fromUContext, pArgs, cnv->invalidUCharBuffer, i, c, reason, pErrorCode);
/* get the converter state from UConverter */
c=cnv->fromUSurrogateLead;
/* update target and deal with offsets if necessary */
offsets=ucnv_updateCallbackOffsets(offsets, ((uint8_t *)pArgs->target)-target, sourceIndex);
target=(uint8_t *)pArgs->target;
/* update the source pointer and index */
sourceIndex+=length+(pArgs->source-source);
source=lastSource=pArgs->source;
targetCapacity=(uint8_t *)pArgs->targetLimit-target;
length=sourceLimit-source;
if(length<targetCapacity) {
targetCapacity=length;
}
/*
* If the callback overflowed the target, then we need to
* stop here with an overflow indication.
*/
if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
break;
} else if(U_FAILURE(*pErrorCode)) {
/* break on error */
c=0;
break;
} else if(cnv->charErrorBufferLength>0) {
/* target is full */
*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
break;
} }
} }
else }
{
*err = U_BUFFER_OVERFLOW_ERROR; if(U_SUCCESS(*pErrorCode) && source<sourceLimit && target>=(uint8_t *)pArgs->targetLimit) {
break; /* target is full */
*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
}
/* set offsets since the start or the last callback */
if(offsets!=NULL) {
size_t count=source-lastSource;
while(count>0) {
*offsets++=sourceIndex++;
--count;
} }
} }
args->target += myTargetIndex; if(pArgs->flush && source>=sourceLimit) {
args->source += mySourceIndex;; /* reset the state for the next conversion */
if(c!=0 && U_SUCCESS(*pErrorCode)) {
return; /* a Unicode code point remains incomplete (only a first surrogate) */
} *pErrorCode=U_TRUNCATED_CHAR_FOUND;
}
U_CFUNC UChar32 T_UConverter_getNextUChar_LATIN_1(UConverterToUnicodeArgs* args, cnv->fromUSurrogateLead=0;
UErrorCode* err) } else {
{ /* set the converter state back into UConverter */
cnv->fromUSurrogateLead=(UChar)c;
/* Empties the internal buffers if need be
* In this case since ErrorFunctors are never called
* (LATIN_1 is a subset of Unicode)
*/
if (args->source+1 > args->sourceLimit)
{
*err = U_INDEX_OUTOFBOUNDS_ERROR;
return 0xffff;
} }
/* make sure that we zero-extend, not sign-extend, the byte */ /* write back the updated pointers */
return (UChar)(uint8_t)*(args->source++); pArgs->source=source;
pArgs->target=(char *)target;
pArgs->offsets=offsets;
} }
static const UConverterImpl _Latin1Impl={ static const UConverterImpl _Latin1Impl={
@ -184,11 +292,11 @@ static const UConverterImpl _Latin1Impl={
NULL, NULL,
NULL, NULL,
T_UConverter_toUnicode_LATIN_1, _Latin1ToUnicodeWithOffsets,
NULL, _Latin1ToUnicodeWithOffsets,
T_UConverter_fromUnicode_LATIN_1, _Latin1FromUnicodeWithOffsets,
NULL, _Latin1FromUnicodeWithOffsets,
T_UConverter_getNextUChar_LATIN_1, _Latin1GetNextUChar,
NULL, NULL,
NULL NULL